The proposed study pertains to aggressive behavior in youth and its relations to biased social information processing. Aggressive behavior is a highly prevalent problem among youth, which carries a high degree of risk for severe short- and long- term negative consequences for the aggressive child and for others. Further, it is one of the most costly problem behaviors in youth. A large body of research shows that social information processing mechanisms are important contributors to aggressive behavior problems in childhood. This study has two major goals. The first is to increase understanding of aggression-relevant biases in children's encoding of social information. Of the various steps in the information-processing sequence, encoding processes are arguably among the most important because they determine what information enters the processing sequence and are thus likely to have substantial influence on processing at all subsequent steps. The second major goal of the study is to increase understanding of differences between two subdimensions of aggressive behavior: Reactive aggression (RA) and proactive aggression (PA), by examining their relations to biased encoding processes. This study builds on a preliminary study showing that RA, but not PA, is associated with a defensive encoding bias in which attention is shifted away from cues associated with threats to self-esteem. The study has 3 specific aims: 1)To replicate this defensive encoding bias and extend understanding of it by relating it to an expanded set of constructs; 2) To refine understanding of this defensive attentional bias by separately measuring preconscious and post-conscious attentional processes. Specifically, it is predicted that this defensive attentional bias occurs only after self-esteem threat cues have entered conscious awareness. It is further predicted that it is preceded by an automatic, preconscious bias toward such threat cues; 3) To test the hypothesis that heightened RA will also be related to an attentional bias toward aggression-relevant physical threat cues but not to aggression-irrelevant physical threat cues. The sample for this study will be 300 children in grades 6-8. To study such attentional biases across the full range of behavior, this sample will include children at all levels of aggression. Previous studies predict that it will include approximately 60 children who show significant levels of aggression. This size sample permits ample power to find expected effects. Child-, parent-, and teacher-report measures of child behavior problems, especially RA and PA, will be collected and children will complete a computer-based task which provides a measure of their allocation of attention to various types of information. Primary study hypotheses will be tested using hierarchical regression analysis.